4 Ethical Codes
What distinguishes ethical codes from other forms of ethics generally is that while they may assign duties and responsibilities, these are assumed voluntarily by virtue of being a member of the profession. To become a nurse is, for example, to adopt as a personal code the ethical norms and values that define that particular profession.
The purpose of this chapter is to showcase the wide range of ethical codes that are employed in different professions, some of which are directly related to the use of analytics in that profession, and others which describe ethics in the profession generally. This diversity is not widely recognized; there is often a presumption, if not an explicit assertion, that the values in these ethical codes, and in ethics generally, are common, core, and universal.
This statement from Metcalf (2014) is typical: “There are several principles that can be found at the core of contemporary ethics codes across many domains:
- respect for persons (autonomy, privacy, informed consent),
- balancing of risk to individuals with benefit to society,
- careful selection of participants,
- independent review of research proposals,
- self-regulating communities of professionals,
- funding dependent on adherence to ethical standards.”
Whether or not one actually believes these principles are foundational, it remains a matter of empirical fact that they are not universal and not core. The same can be said for similar assertions of universality made elsewhere (for example: Pitofsky, (1998:7), Singer & Vinson (2002), CPA (2017)).
This chapter is a substantial survey of dozens of ethical codes. Though every attempt has been to keep this treatment brief, it is nonetheless not brief. By laying out the evidence I endeavour to show, rather than argue, that there is no common foundation to the ethical codes that govern different professions.
We’ll begin with a quick overview of what we mean by ethical codes, discussing the purpose and operation of ethical codes, some of the components of ethical codes, and the ways in which these codes differ from each other. Then we’ll take an extended look at the issues raised by the codes. First we look at what problems the codes are trying to solve, or in other words, what the purpose was for writing the codes. Then we look at a length list of values and priorities revealed in the codes. After this examination, we consider the question, to whom are the professionals described in the codes obligated? Finally, we ask what bases and foundations underlie the recommendations in the codes.
The full set of ethical codes is displayed, with readers invited to notice the ways in which they differ from each other, in Appendix 1: An Ethical Codes Reader, with references linking back to the full code in question, for further study as desired by the reader.
Standards of Conduct
Why Ethical Codes?
The need for professional ethics encompasses a number of factors. There is the need to be able to trust a person in a position of trust. There is the need to make good decisions and to do the right thing. And then there are various intangibles. The Project Management Institute (PMI, 2020) states, “Ethics is about making the best possible decisions concerning people, resources and the environment. Ethical choices diminish risk, advance positive results, increase trust, determine long term success and build reputations. Leadership is absolutely dependent on ethical choices.”
But these are not the only reasons advanced to justify professional ethics. There is the concern that without a statement of ethics, unethical conduct will abound. “The absence of a formal code could be seen almost as a guarantee that if such cases did exist they would be swept under the carpet, left to others (probably the law) to sort out,” writes Sturges (2003).
Others are less concerned about good behaviour per se than they are about the bottom line. Alankar Karpe (2015), for example, writes in ‘Being Ethical is Profitable’ that “Shortcuts and sleazy behavior sometimes pay handsomely, but only for the short term. Organizations must remember that any benefits from lying, cheating, and stealing usually come at the expense of their reputation, brand image, and shareholders.” And, as he notes, ““There is one and only one social responsibility of business – to use it[s]resources and engage in activities designed to increase its profits so long as it stays within the rules of the game, which is to say, engages in open and free competition.”
Additionally, there are services and institutions that require professional ethics in order to function. For example, the CFA Institute (2017) states, “ethical conduct is vital to the ongoing viability of the capital markets.” It notes, “compliance with regulation alone is insufficient to fully earn investor trust. Individuals and firms must develop a ‘culture of integrity’ that permeates all levels of operations.” Indeed, it is arguable that society as a whole could not function without professional ethics. Thus, the “CFA Institute recently added the concept ‘for the ultimate benefit of society’ to its mission.”
Certain disciplines see ethical codes as essential to being recognized as a profession. Hence, for example, for librarians, “Keith Lawry set the idea of a code in a particularly positive view of the professionalization process in British librarianship. He linked the Library Association’s possession of a code of professional conduct with the potential for statutory recognition of the association’s control of who might and who might not practise librarianship” (Sturges, 2003)
Finally, practitioners need them. As Rumman Chowdhury, Accenture’s Responsible AI Lead, said, “I’ve seen many ‘ethics codes’ focused on AI, and while many of them are very good they’re more directional than prescriptive – more in the spirit of the Hippocratic Oath that doctors are expected to live by. Meanwhile, many data scientists are hungry for something more specific and technical. That’s what we need to be moving toward” (De Bruijn, et.al., 2019)
Ethical Codes As Standards of Conduct
While ethics commonly applies to people in general, there is a specific class of ethics that applies to people by virtue of their membership in a professional group. There are different approaches, but in general, “professional ethics are principles that govern the behaviour of a person or group in a business environment. Like values, professional ethics provide rules on how a person should act towards other people and institutions in such an environment” (Government of New Zealand, 2018).
Professional ethics can be characterized as imposing a higher standard of conduct. The reasons for this vary, but (as we discuss below) a higher standard is demanded because professionals are in positions of power, they have people in their care, and they are expected to have special competencies and responsibilities. Additionally, professional ethics may require that practitioners put the interests of others ahead of their own. This may include duties not only to those in one’s care, but also to clients, organizations, or even intangibles like ‘the Constitution’ or ‘the public good’.
As such, professional ethics are often expressed in terms of codes of conduct (indeed, it is hard to find a sense of professional ethics where such a code is not employed). Though the code is normative (“breaches of a code of conduct usually do carry a professional disciplinary consequence” (Ibid.)) usually the intent of the code is to remind professionals of their duty and prompt them regarding specific obligations.
Ethical Codes as Requirements
In the world of software engineering, in addition to ethical standards as codes of conduct, ethical codes can be seen as defining requirements. This is proposed, for example, by Guizzardi, et.al. (2020). They write, “Ethical requirements are requirements for AI systems derived from ethical principles or ethical codes (norms). They are akin to Legal Requirements, i.e.,requirements derived from laws and regulations.” Ethical requirements are drawn from stakeholders in the form of principles and codes. From these, specific requirement statements are derived. “For example, from the Principle of Autonomy one may derive “Respect for a person’s privacy”, and from that an ethical requirement “Take a photo of someone only after her consent” (Ibid: 252).
An important distinction between the idea of ethical codes as standards of conduct and ethical codes as requirements is that in the former case, the AI is treated as an ethical agent can reason and act on the basis of ethical principle, while in the latter case, the AI is not. “Rather, they are software systems that have the functionality and qualities to meet ethical requirements, in addition to other requirements they are meant to fulfill” (Ibid: 252).
As Opposed to Legal Requirements
We stated above that ‘ethics is not the same as the law’. This is a case where that principle applies. What we are interested in here is the sense of an ethical code as a principle of ethics, not as a legal document. It reflects the fact that a person chooses a profession for themselves, and thereby voluntarily enters into a set of obligations characterized by that profession. “Professions must be ‘professed’ (that is, declared or claimed)” (Davis, 2010:232).
Thus we may say that ethics may be influenced by, but are distinct from, the following (all from Government of New Zealand, 2018):
- Fiduciary duties – fiduciary duties are “special obligations between one party, often with power or the ability to exercise discretion that impacts on the other party, who may be vulnerable” (Wagner Sidlofsky, 2020). Examples of fiduciary relations include those between lawyer and client, trustee and beneficiary, director and company, power of attorney and beneficiary and accountant and client.
- Contractual obligations – these require the professional to perform the terms of the contract, and “includes a duty to act with diligence, due care and skill, and also implies obligations such as confidentiality and honesty” (New Zealand, 2018).
- Other laws – for example, In New Zealand this could include the Consumer Guarantees Act 1993.
What distinguishes legal requirements, arguably, from ethical principles is the element of choice. In the case of legal requirements, the law compels you to behave in a certain way, with increasing penalties for non-compliance. In an important sense, it doesn’t matter whether the law or the principle in question is ethical or not. You are penalized if you do not comply.
It may be argued that the relation between ethics and law is such that in a treatment of the ethics of learning analytics we ought also to be concerned with the law in relation to learning analytics. We will see this come up in two ways: first, in the argument that ‘obeying the law’ is part of the ethical responsibility of a practitioner, and second, in the argument that the law regarding learning analytics is or ought to be informed by ethical principles.
Principles and Values
“Values are general moral obligations while principles are the ethical conditions or behaviors we expect” (Gilman, 2005: 10). Values and principles are connected. As Terry Cooper (1998:12) explains, “An ethical principle is a statement concerning the conduct or state of being that is required for the fulfillment of a value; it explicitly links a value with a general mode of action.” For example, we may state that we value ‘justice’, but we would need a principle like “treat equals equally and unequals unequally” to explain what we mean by ‘justice’.
All ethics codes encompass both principles and values, though (as we shall see below) usually more implicitly than explicitly. Values (such as honesty and trustworthiness) are often assumed tacitly, as not needing to be stated. Sometimes they are expressed in a preamble to the code, not as an explicit list, but rather in the sense of establishing a context. For example, the Canadian Code of Public Service ethical code has a preamble describing the role of the public service, as well as a listing of the fundamental values (TBS, 2011).
The Value of Professional Codes
Codes of professional ethics or conduct are widely used. They bring a utilitarian value to the conversation. They provide a framework for professionals carrying out their responsibilities. They clearly articulate unacceptable conduct. And they provide a vision toward which a professional may be striving (Gilman, 2005: 5) Having a code, it is argued, is key to the prevention of unacceptable conduct. That’s why, for example, the United Nations Convention Against Corruption included a public service code of conduct as an essential element in corruption prevention, says Gilman (Ibid). Yet the convention is an interesting example: there is no code of conduct for the private sector. Why?
At the same time, it is argued that “Codes are not designed for ‘bad’ people, but for the persons who want to act ethically” (Ibid: 7). That is, they provide guidance for a person who wants to act ethically, but who may not know what is right. Therefore, codes are preventative only in the sense that they prevent conduct that is accidentally unacceptable. They may seem to be unnecessary in the case of a well-developed profession and body of professionals, but in a new environment, such as data analytics in education, there is much that is not yet clearly and widely understood.
Moreover, argues Gilman, a code of ethics will change the behaviour of bad actors, even if it does not incline them toward good. “When everyone clearly knows the ethical standards of an organization they are more likely to recognize wrongdoing; and do something about it. Second, miscreants are often hesitant to commit an unethical act if they believe that everyone else around them knows it is wrong. And, finally corrupt individuals believe that they are more likely to get caught in environments that emphasize ethical behavior.” (Ibid: 8)
Study of Ethical Codes
More than 70 ethical codes were studied as a part of this review. The selection methodology undertaken was designed to encourage as wide a range of ethical codes as possible. To begin, ethical codes referenced in relevant metastudies (such as ) were evaluated. Codes referenced by these ethical codes were studied, to establish a history of code development within a discipline. Documents from relevant disciplinary associations were studied, to find more ethical codes. The selection of ethical codes includes the following major disciplinary groups (and the number of individual codes studied).
- Professional ethics – broad-based ethical codes (4)
- Academic ethics – codes of conduct for professors and staff in traditional academic institutions (3)
- Teacher ethics – codes governing teachers and the teaching profession (7)
- Ethics for librarians and information workers – ethics of information management (2)
- Public service ethics – codes of conduct for government employees (2)
- Research ethics – includes international declarations and government policy (6)
- Health care ethics – including codes for doctors and nurses (6)
- Ethics in social science research – research ethics (1)
- Data ethics – government and industry declarations on the use of study and survey data (7)
- Market research ethics – codes describing the ethical use of data in advertising and market studies (2)
- Journalism ethics – codes of conduct governing the use of public information by journalists (3)
- Ethics for IT professionals – system administration and software development ethics (3)
- Data research ethics – related specifically to the use of data in research (1)
- Ethics for artificial intelligence – government, industry and academic codes (15)
- Information and privacy – principles specifically addressing individual rights (1)
- Ethics in educational research – policies governing educational researchers specifically (3)
- Ethics in learning analytics – government, academic and industry guidelines and codes (7)
Who Writes the Codes
As van Nuland & Khandelwal (2006:18) write in relation to teacher ethics, “In some places, they were developed by authorities in charge of the public sector, such as the Ministry of Education (Bangladesh, India, Nepal); in others, they are designed by an autonomous body (Hong Kong) or by teacher organizations themselves (the Province of Ontario in Canada). “
(( Section to be completed ))
How the Codes Differ
Metcalf (2014) identifies a number of the reasons ethical codes vary across professions, and even within professions (quotes in the list below are all from Metcalf):
- Motivation: The events that prompt the development of ethical codes; for example, “in biomedicine, ethics codes and policies have tended to follow scandals” while by contrast “major policies in computing ethics have presaged many of the issues that are now experienced as more urgent in the context of big data.”
- Purpose: “Analyses of ethics codes note a wide range of purposes for ethics codes (Frankel, 1998; Gaumintz and Lere, 2002; Kaptein and Wempe, 1998).”
- Interests: “Frankel (1989) notes that all ethics codes serve multiple interests and therefore have multiple, sometimes conflicting, dimensions. He offers a taxonomy of aspirational, educational, and regulatory codes.”
- Burden: who does the ethical code apply to? Metcalf notes that “greater burdens are placed on individual members to carry out the profession’s ethical agenda,” but different burdens may fall on different groups of people.
- Enforcement: “Organizations, institutions and communities tend to develop methods of enforcement that reflect their mission.”
Each code of ethics was subjected to an analysis that includes the following criteria:
- What ethical issues is it attempting to address (for example, is focused on malpractice, on conflict of interest, on violation of individual rights, etc)?
- What are its core values or highest priorities (as opposed to the detailed specification of ethical principles described, as defined by Cooper (1998:12), Gilman (2005: 10))?
- Which ethical issues from the literature of learning analytics issues do they address?
- Who is governed, and to whom are they obligated? (e.g.,AITP (2017) list six separate groups to which information professionals have obligations).
- What is the basis (if any) for the statement of ethical values and principles (for example, the Royal Society’s recommendations are based in a “public consultation” (Drew, 2018)), while numerous other statements are based in principles such as ‘fairness’ and ‘do no harm’.
Focus on Ethical Issues
In this section we examine the ethical issues being addressed by codes of conduct. Most often these are not stated explicitly, but must be inferred from the sorts of behaviours or outcomes being expressly discussed.
The Good that Can Be Done
While ethical codes are typically thought of as identifying wrongs, in the sense of “thou shalt not”, it should be noted that many codes reference first the good that can be accomplished by the discipline or profession being discussed. This is especially the case in relation to data management and data research, which are new fields, and where the benefits may not be immediately obvious.
For example, while the United Kingdom Data Ethics Framework “sets out clear principles for how data should be used in the public sector,” it is with the intention to “maximise the value of data whilst also setting the highest standards for transparency and accountability when building or buying new data technology” (Gov.UK, 2018), advising researchers to “start with clear user need and public benefit.” Also in the U.K., the list of principles outlines by the House of Lords Select Committee on AI principles reflect a purpose “for the common good and benefit of humanity” including privacy rights, the right to be educated, “to flourish mentally, emotionally and economically alongside artificial intelligence” (Clement-Jones, et.al, 2018, para 417).
Similarly, the Sorbonne Declaration (2020) points to “the benefit of society and economic development” that accrues as a research of data research. It is motivated by the good that can be done and “recognises the importance of sharing data in solving global concerns – for example, curing diseases, creating renewable energy sources, or understanding climate change” (Merett, 2020). In some cases, the emphasis is on being able to be more ethical. The Society of Actuaries, “AI provides many new opportunities for ethical issues in practice beyond current practices,” for example, ‘black box’ decision models, masked bias, and unregulated data” (Raden, 2019: 9), all issues that received much less attention in the days before analytics.
In the field of learning analytics, there is often an explicit linkage drawn between the use of data and benefits for students, and thereby, of helping society benefit from education generally. The Open University, for example, asserts that the purpose of collecting data should be “to identify ways of effectively supporting students to achieve their declared study goals” (OU, 2014:4.2.2). The Asilomar Convention for Learning Research in Higher Education principles were based on “the promise of education to improve the human condition”, as expressed by two tenets of educational research: to “advance the science of learning for the improvement of higher education”, and to share “data, discovery, and technology among a community of researchers and educational organizations” (Stevens & Silbey, 2014).
Academic or Professional Freedom
Ethical codes frequently point to the need for freedom or autonomy for the profession. Not surprisingly, the concept of academic freedom surfaces frequently in academic codes of ethics. It is seen as something that needs to be nurtured and protected. Thus, for example, one university’s code of ethics asserts that the defense of academic freedom is an “obligation” on faculty members, stating, “it is unethical for faculty members to enter into any agreement that infringes their freedom to publish the results of research conducted within the University precincts or under University auspices… they have the obligation to defend the right of their colleagues to academic freedom. It is unethical to act so as deliberately to infringe that freedom” (SFU, 1992). Or, good practices are those that defend academic freedom (EUI, 2019).
But university professors are not along in asserting professional independence. Researchers generally, and especially early-career researchers (ECR) “are being pressured into publishing against their ethics because of threats relating to job security” (Folan, 2020). Librarians declare that they are “explicitly committed to intellectual freedom and the freedom of access to information. We have a special obligation to ensure the free flow of information and ideas to present and future generations” (ALA, 2008). Doctors and nurses also declare the caregiver’s right to “be free to choose whom to serve, with whom to associate, and the environment in which to provide medical care” (AMA, 2001). The same assertions of independence and autonomy can be found in journalists’ code of ethics (NUJ, 2011).
Conflict of interest
The idea that a person would use their position to personally benefit from their position of privilege or responsibility, whether directly or through the offer of gifts or benefits, is expressly prohibited by many (but by no mean all) codes of ethics (CFA, 2019; IEEE, 2020: 7.8; SFU, 1992; CPA, 2017). Different sorts of conflict of interest are mentioned by different codes of ethics.
Some codes focus on material benefits. For example, codes of ethics in the financial sector often express prohibitions against insider trading (specifically, members that “possess material nonpublic information that could affect the value of an investment must not act or cause others to act on the information” and against “practices that distort prices or artificially inflate trading volume with the intent to mislead market participants” (CFA, 2019). Public services ethics., meanwhile, address conflict of interest as a matter of trust where the principles include “taking all possible steps to prevent and resolve any real, apparent or potential conflicts of interest,” as well as “effectively and efficiently using the public money, property and resources managed by them” (TBS 2011).
Other codes focus on integrity. We see this in professions like journalism, where “professional integrity is the cornerstone of a journalist’s credibility” (SPJ, 1996) and journalists are urged “to remain independent (and therefore avoid conflict of interest), and to be accountable” (SPJ, 2014). The primary focus of the New York Times Ethical Journalism Guidebook is avoidance of conflict of interest, and it addresses exhaustively the ways in which a journalist could be in a real or perceived conflict of interest, and counsels against them, while allowing for certain exceptions (NYT, 2018).
In education and the helping professions the codes focus on exploitation (IUPSYS, 2008; CPA, 2017; NEA, 1975; BACB, 2014:6; SFU, 1992; EUI, 2019 etc.). The British Columbia Teachers Federation, for example, states that “a privileged relationship exists between members and students” and stresses the importance of refraining from exploiting that relationship” (BCTF, 2020).
The prevention of harm is a theme that arises in numerous codes of ethics. Many codes trace their origins to the written principles for ethical research originating from the Nuremberg trials in 1949 that were used to convict leading Nazi medics for their atrocities during the Second World War (Kay et al. 2012). In general, research should not risk “even remote possibilities of injury, disability, or death,” nor should the harm exceed the potential benefits of the research (USHM, 2020). What counts as harm, however, varies from code to code.
Often, the nature of harm is loosely defined. Accenture’s Universal Principles for Data Ethics (Accenture, 2016:5) states that practitioners need to be aware of the harm the data could cause, both directly, and through the “downstream use” of data. The principles also acknowledge that data is not neutral. “There is no such thing as raw data.” The Information Technology Industry Council urges researchers to “Recognize potentials for use and misuse, the implications of such actions, and the responsibility and opportunity to take steps to avoid the reasonably predictable misuse of this technology by committing to ethics by design. (UC Berkeley, 2019)
Discrimination and human rights violations are often cited as sources of harm (IEEE,2020: 9.26; NEA, 1975; IFLA, 2012; NUJ, 2011; UC Berkeley, 2019; etc.). For example, the Amnesty International and Access Now ‘Toronto Declaration’ calls on the right to redress human rights violations caused by analytics and AI. “This may include, for example, creating clear, independent, and visible processes for redress following adverse individual or societal effects,” the declaration suggests, “[and making decisions] subject to accessible and effective appeal and judicial review” (Brandom, 2018).
Several codes, by contrast, identify exemptions and cases that will not be considered harm. For example, the U.S. ‘Common Rule’ states that research is exempt from restrictions if it is a “benign behavioral exemption”, that is, it is “brief in duration, harmless, painless, not physically invasive, not likely to have a significant adverse lasting impact on the subjects, and the investigator has no reason to think the subjects will find the interventions offensive or embarrassing” (HHS, 2018:§46.104.2.C.ii).
Quality and Standards
Ethical codes – especially professional ethical codes – also address issues related to quality and standards. Sometimes competence is defined simply as “stewardship and excellence” (TBS,2011)or professionalism (CFA, 2019; BACB, 2014:6). Or a profession may seek to restrict practice to competent practitioners, for example, preventing assistance to a “noneducator in the unauthorized practice of teaching” and preventing “any entry into the profession of a person known to be unqualified in respect to character, education, or other relevant attribute” (NEA, 1975).
The code may also seek to define and reinforce exemplary behaviours such as research integrity, scientific rigor and recognition of sources. The ethical code for behavioural analysts, for example, states that researchers must not fabricate data or falsify results in their publications, must correct errors in their publications, and not omit findings that might alter interpretations of their work (BACB,2014:9.0). Similarly, “The IEEE acknowledges the idea of scientific rigor in its call for creators of AI systems to define metrics, make them accessible, and measure systems” (Feljd, et.al., 2020:59). The major sources of academic misconduct are related to the misuse of intellectual property, for example, through plagiarism, piracy, misrepresentation of authorship (“personation”), and fabrication data or qualifications (EUI, 2019; BACB,2014:9.0).
What are the Limits?
Finally, some ethical codes seek to address the limits of what can be done ethically. It’s not always easy to recognize these limits; it was only after years of effort that IBM announced it would see work in general facial recognition technology, for example (Krishna, 2020). Sometimes the need for limits is stated explicitly. The purpose of the U.K. Government Data Ethics Framework, for example, to help data scientists identify the limits of what is allowed, to help practitioners consider policy when designing data science initiatives, and to identify core ethical expectations from such projects (Gov.UK, 2018).
Some discussions (eg. Floridi, et.al., 2018, note 5) omit consideration of the research issues (arguing “they are related specifically to the practicalities of AI development”), however they set an important ethical standard, specifically, “to create not undirected intelligence, but beneficial intelligence” (Asilomar, 2017). In other cases, specific outcomes are undesired, for example, “We should not build a society where humans are overly dependent on AI or where AI is used to control human behavior through the excessive pursuit of efficiency and convenience” (Japan, 2019:4). Many individual researchers, meanwhile, refuse to work on military or intelligence applications (Shane & Wakabayashi, 2018).
Otherwise, the limits are related to the benefits. For example, the Information and Privacy Commissioner Ontario, Canada. Data-gathering by the state should be restricted to that which is reasonably necessary to meet legitimate social objectives, and subjected to controls over its retention, subsequent use, and disclosure. (Cavoukian, 2013). Similarly, research Ethics Boards (REB) often require that the submissions for ethics approval be accompanied with statements of scientific merit and research need.
Core Values and Priorities
The previous section addressed ethical issues being addressed by codes of conduct. It was, in a sense, addressing the purpose of the code qua code of ethics, that is, it didn’t look at the social, political or economic need for codes of ethics, but rather, sought to identify the questions for which a ‘code of ethics’ is the answer. No code of those surveyed was designed to meet all of the purposes identified, and none of the purposes identified was specifically addressed by all of the codes surveyed. We use different ethical codes to do different things.
In this section, we will focus on the values and priorities that can be found in the codes. These are things that might be found in the ethical principles described by the code, if the code is structured that way, or the things that are explicitly described as good or desirable by the code. When people state that there is a ‘universal’ or ‘general’ agreement on values, it is usually with respect to a subset of the items listed here that they refer. Below we have not attempted to create a tab or values mapped to codes, as some researchers (eg. Fjeld, et.al., 2020) have done, but rather, to list the values with references to relevant examples where they are asserted.
Pursuit of Knowledge
The pursuit of knowledge is identified as a core value by many academic and professional codes. For example, the SFU code of ethics, addresses faculty members first as teachers, and then as scholars. “The first responsibility of university teachers is the pursuit and dissemination of knowledge and understanding through teaching and research. They must devote their energies conscientiously to develop their scholarly competence and effectiveness as teachers” (SFU, 1992).
Similarly, the National Education Association statement (NEA, 1975) is based on “recognizes the supreme importance of the pursuit of truth, devotion to excellence, and the nurture of the democratic principles.” Nor is the pursuit of knowledge limited to academics. The Society for Professional Journalists (SPJ) code of ethics, originally derived from Sigma Delta Chi’s ‘New Code of Ethics’ in 1926 (SPJ, 2014), asserts that the primary function of journalism, according to the statements, is to inform the public and to serve the truth.
Autonomy and Individual Value
Many codes, like National Education Association code (NEA, 1975) are based on “believing in the worth and dignity of each human being. This, though, is expressed in different ways by different codes. For example, in one code, individual development is the objective, to promote “acquisition of autonomous attitudes and behavior.” (Soleil, 1923). The AI4People (Floridi, et.al., 2018:16) adopts a similar stance.
By contrast Tom Beauchamp and James Childress’s Principles of Biomedical Ethics contains an extended discussion of autonomy embracing the idea of ‘informed consent’, which requires disclosure of information, respect for decision-making, and provision of advice where requested. A similar respect for human autonomy is demanded by the High-Level Expert Group on Artificial Intelligence (AI HLEG, 2019).
Similarly, the Belmont Report begins by identifying ‘respect for persons’, as a core principle which “incorporates at least two basic ethical convictions: first, that individuals should be treated as autonomous agents, and second, that persons with diminished autonomy are entitled to protection.” (DHEW, 1978:4)
Whether or not based in the principle of autonomy or the inherent worth of people, the principle of consent is itself often cited as a fundamental value by many ethical codes (BACB, 2014; DHEW, 1978; HHS, 2018; Drachsler & Greller, 2016, etc.). However there may be variations in what counts as consent and what consent allows.
For example, the type of consent defined by the Nuremberg Code “requires that before the acceptance of an affirmative decision by the experimental subject there should be made known to him the nature, duration, and purpose of the experiment; the method and means by which it is to be conducted; all inconveniences and hazards reasonably to be expected; and the effects upon his health or person which may possibly come from his participation in the experiment” (USHM, 2020).
Several codes are more explicit about what counts as informed consent. For example, one code requires that “researchers be transparent about the research and give research subjects the choice not to participate. This includes passive data collection, such as collection of data by observing, measuring, or recording a data subject’s actions or behaviour” (IA, 2019). The same code, however, contains provisions that allow data to be collected without consent. If consent is not possible, it states, “Researchers must have legally permissible grounds to collect the data and must remove or obscure any identifying characteristics as soon as operationally possible.” There are also stipulations designed to ensure research quality and to ensure that communications about the research are accurate and not misleading (Ibid).
Meanwhile, that same code of ethics can allow the scope of consent to be extended beyond research. It is the IA Code of Standards and Ethics for Marketing Research and Data Analytics (IA, 2019). Consent is required for research purposes, but in addition “such consent can enable non-research activities to utilize research techniques for certain types of customer satisfaction, user, employee and other experience activities.” The Nuremberg Code and marketing research may stand at opposite poles of an ethical question, however, they are reflective of a society as a whole that holds consent as sacrosanct on one hand and makes legal End User Licensing Agreements (EULA) on the other hand.
Integrity is often required of professionals (CFA, 2019; CSPL, 1995; IA, 2019; etc.), but different codes stress different aspects of integrity. The Canadian Psychological Association section on integrity speaks to accuracy, honesty, objectivity, openness, disclosure, and avoidance of conflict of interest (CPA, 2017). The European University Institute defines integrity as including such values as honesty, trust, fairness and respect. (EUI, 2019). The Ontario College of Teachers focuses on trust, which includes “fairness, openness and honesty” and integrity, which includes honesty and reliability (OCT, 2020). In Guyana, integrity includes “honest representation of one’s own credentials, fulfilment of contracts, and accountability for expenses” (Guyana, 2017). The Nolan Principles state “Holders of public office should act solely in terms of the public interest” (CSPL, 1995) while Raden (2019: 9) defines it as “incorruptibility”.
While sometimes breaches of confidentiality are depicted as ‘harm’, confidentiality is often presented as a virtue in and of itself, perhaps constitutive of integrity. Thus, for example, librarians “protect each library user’s right to privacy and confidentiality with respect to information sought or received and resources consulted, borrowed, acquired or transmitted” (ALA, 2008). Similarly, the Declaration of Helsinki states that “every precaution must be taken to protect the privacy of research subjects and the confidentiality of their personal information” (WMA, 2013).
The need for confidentiality increases with the use of electronic data. The authors of a 1973 report for the U.S. Department of Health, Education and Welfare addressing the then nascent practice of electronic data management noted that “under current law, a person’s privacy is poorly protected against arbitrary or abusive record-keeping practices” (Ware, et.al., 1973:xx). Government policy, they argued, should be designed to limit intrusiveness, to maximize fairness, and to create legitimate and enforceable expectations of confidentiality (Linowes, et.al.,1977: 14-15).
Confidentiality, expressed as privacy, is a core principle for data and information services and codes regulating those. For example, the Federal Trade Commission promotes principles that “are widely accepted as essential to ensuring that the collection, use, and dissemination of personal information are conducted fairly and in a manner consistent with consumer privacy interests.” (Pitofsky, et.al., 1998:ii).
It should be noted that exceptions to confidentiality may be allowed, especially where required by law. For example, the British Columbia Teachers’ Federation code states explicitly that “It shall not be considered a breach of the Code of Ethics for a member to follow the legal requirements for reporting child protection issues” (BCTF, 2020). Similarly, in medical informatics, confidentiality can be compromised “by the legitimate, appropriate and relevant data-needs of a free, responsible and democratic society, and by the equal and competing rights of others” (IMIA, 2015).
Care, which includes “compassion, acceptance, interest and insight for developing students’ potential” (OCT, 2020) is found in numerous ethical codes (CNA, 2017; CFA, 2019; IUPSYS, 2008; CPA, 2017; etc.) but is manifest differently in each code in this it appears. Contrasting the OCT definition, for example, is the Canadian Nurses Association discussion of “provision of care” references speech and body language, building relationships, learning from “near misses”, adjusting priorities and minimizing harm, safeguarding care during job actions, and more. It is worth noting that the promotion of dignity means to “take into account their values, customs and spiritual beliefs, as well as their social and economic circumstances without judgment or bias.” (CAN, 2017:12)
The National Council of Educational Research and Training is almost unique in an assertion of care, in the explanatory notes, that states “the demonstration of genuine love and affection by teachers for their students is essential for learning to happen. Treating all children with love and affection irrespective of their school performance and achievement level is the core of the teaching learning process” (NCERT, 2010).
Other codes (eg. CFA, 2019) adopt a more legalist interpretation of ‘duty of care’, for example, that researchers must “prioritize data subject privacy above business objectives, be honest, transparent, and straightforward in all interactions (and respect the rights and well-being of data subjects” (IA, 2019). Meanwhile there is a sense of ‘care’ that means ‘diligence and rigor’; this is the sense intended in the Nuremberg Code (USHM, 2020) and the American Medical Association (Riddick, 2003).
Competence and Authority
Many of the codes identify competence or authority to practice in the profession as core values or principles (CFA, 2019; IEEE, 2020: 7.8; IUPSYS, 2008; etc.). This is expressed in several ways: members of the profession may be expected to perform in a competent manner, or they may be required to remain within their domain of competence, or they may be obligated to ensure that unqualified people do not practice the profession (NEA, 1975, as cited above).
For example, behaviour analysts are expected to rely on scientific evidence and remain within the domain of their competence (BACB, 2014:6). Similarly, the Nuremberg Code also determines that the researcher should be a qualified scientist and that the research ought to have scientific merit and be based on sound theory and previous testing (USHM, 2020). And the CPA code (2017) requires that the practitioner be competent.
Sometimes what counts as competence is spelled out in the code. For example, the Royal Society data science ethics in government report (Drew, 2016) advises the use of robust data models in data research. Provisions in the Open University code similarly state that the modeling based on the data should be sound and free from bias, and that it requires “development of appropriate skills across the organisation” (OU, 2014:4.4).
Codes sometimes require that only authorized professionals perform the work. Accenture’s Universal Principles for Data Ethics (Accenture, 2016:5) states “practitioners should accurately represent their qualifications (and limits to their expertise).” This is especially the case where expertise is more difficult to establish or where the stakes are higher. The Guyana code of ethics for teachers, for example, requires “honest representation of one’s own credentials” (Guyana, 2017) while the Ontario Information and Privacy Commissioner Ontario states that “the authority to employ intrusive surveillance powers should generally be restricted to limited classes of individuals such as police officers” (Cavoukian, 2013).
Value and Benefit
While above we represented ‘the good that can be done’ as aspirational, that is, something ethical codes seek to accomplish, in the present case we view the same principle as a limit, and specifically, as the research or practice must produce a benefit in order to be ethical.
In some cases, this benefit may be immediate and practical. For example, the Behavior Analyst Certification Board requires that practitioners provide “effective treatment” (BACB, 2014:6). It is arguable, as well, that “health-care professionals, especially, have an obligation to distinguish between remedies that represent the careful consensus of highly trained experts and snake oil” (Kennedy, et.al., 2002).
In other cases the requirements are more general (and more widely distributed). The Royal Society requires that researchers “show clear user need and public beneﬁt” (Drew, 2016). Similarly, the Asilomar principles state that “AI technologies should benefit and empower as many people as possible” and “the economic prosperity created by AI should be shared broadly, to benefit all of humanity” (Asilomar, 2017). Fjeld (2020) finds a principle of “promotion of human values,” and specifically, that “the ends to which AI is devoted and the means by which it is implemented should promote humanity’s well being.”
In other cases, the requirement that a benefit be shown is limited to requiring that practitioners demonstrate a purpose for their work. The Barcelona Principles (2010) for example require that researchers “specify purposes of data gathering in advance, and seek approval for any new uses,” while the DELICATE principles require that universities “Decide on the purpose of learning analytics for your institution” and “E-xplain: Define the scope of data collection and usage” (Drachsler & Greller, 2016).
The principle of non-maleficence is an adaptation of the principle of “do no harm” in the Hippocratic oath. This adaptation is necessary because harm is unavoidable in many circumstances; the surgeon must sometimes harm in order to heal, for example. Harm may occur in other professions as well; a teacher might punish, a researcher might violate privacy, a defence contractor might develop weapons.
So the principle of non-maleficience, as developed for example by Beauchamp & Childress (1992) means “avoiding anything which is unnecessarily or unjustifiably harmful… (and) whether the level of harm is proportionate to the good it might achieve and whether there are other procedures that might achieve the same result without causing as much harm” (Ethics Centre, 2017). The principle arguably also requires consideration of what the subject considers to be harm because as Englehardt (1993) says, we engage one another as moral strangers who need to negotiate moral arrangements (Erlanger, 2002).
The definition of maleficence to be avoided can be variably broad. For example, the AMA (2001) addresses not only the nature and priority of patient care, but also “respect for law, respect of a patient’s rights, including confidences and privacy.” The AMA’s Declaration of Professional Responsibility also advocates “a commitment to respect human life” which includes a provision to “refrain from crimes against humanity (Riddick, 2003).
The principle of non-maleficence is found in numerous ethical codes, and not only medical ethics. For example, the Association for Computing Machinery (2018) states “an essential aim of computing professionals is to minimize negative consequences of computing, including threats to health, safety, personal security, and privacy,” including “examples of harm include unjustified physical or mental injury, unjustified destruction or disclosure of information, and unjustified damage to property, reputation, and the environment” (ACM, 2018).
Non-maleficence in research and data science include being minimally intrusive (Drew, 2016), to keep data secure (ibid; also Raden, 2019: 9), to promote “resilience to attack and security, fall back plan and general safety, accuracy, reliability and reproducibility… including respect for privacy, quality and integrity of data, and access to data” (AI HLEG, 2019). AI systems, says Fjeld (2020) should perform as intended and be secure from compromise (also Drachsler & Greller, 2016).
Another of the principles defined by Beauchamp & Childress (1992), beneficence should be understood as more than non-maleficence and distinct from value and benefit. A professional demonstrates beneficence toward their client “not only by respecting their decisions and protecting them from harm, but also by making efforts to secure their well-being.” Moreover, “beneficence is understood in a stronger sense, as an obligation.” It’s intended as a combination of “do no harm” and “maximize benefits and minimize harm”, with the recognition that even the determination of what is harmful might create a risk of harm (DHEW, 1978:6-7).
In a number of ethical codes, beneficence can be thought of as “the principle of acting with the best interest of the other in mind” (Aldcroft, 2012). This is more than merely the idea of doing good for someone, it is the idea that the role of the professional is to prioritize the best interest of their client (BACB, 2015; AMA, 2001; CPA, 2017). The principle of beneficence is also raised with respect to AI (Floridi, et.al, 2018:16; Stevens & Silbey, 2014), however, in the precise statement of these principles it is unclear how they should be applied. For example, should ‘the common good’ is included in the principle of beneficence? Should AI promote social justice, or merely be developed consistently with the principles of social justice?
The principle of respect is cited in numerous ethical codes (AMA, 2001; IUPSYS, 2008; CPA, 2017; Dingwell, et.al., 2017; etc.), for example, acting toward students with respect and dignity (BCTF, 2020), “respect for people” (TBS, 2011), “mutual respect” (Folan, 2020), “respect for the composite culture of India among students” (NCERT, 2010), or “respect for the rights and dignity of learners” (Stevens & Silbey, 2014). Though sometimes paired with autonomy (DHEW, 1978:4, cited above) it is often presented quite differently. The Ontario College of Teachers code states that respect includes trust, fairness, social justice, freedom, and democracy (OCT, 2020).
Respect can also be thought of as promoting “human dignity and flourishing”, which AI4All summarizes as “who we can become (autonomous self-realisation); what we can do (human agency); what we can achieve (individual and societal capabilities); and how we can interact with each other and the world (societal cohesion)” (Floridi, et.al., 2018:7). The last two ‘commandments’ of the Computer Ethics Institute’s Ten Commandments of Computer Ethics recommend computer professionals “think about the social consequences” and to “ensure consideration and respect for other humans” (CEI, 1992).
Several ethical codes include ‘respect for democracy’ among their values and principles; this can mean, variously, respect for the idea of rule by the people, respect for the results of democratic choice (as, say, found in public service ethics; TBS,2011:1.1-1.2), and respect for democratic values, such as justice and non-discrimination.
Democracy is also identified as both an input and output of ethical codes; the NEA code (1975) is based on “the nurture of the democratic principles,” while the Code of Professional Ethics for School Teachers in India states that “every child has a fundamental right to receive education of good quality,” where this education develops the individual personality, faith in democracy and social justice, cultural heritage and national consciousness (NCERT, 2010).
Justice and Fairness
Almost all the ethical codes consulted refer to justice in one form or another. Here it is listed alongside ‘fairness’, as ever since John Rawls’s influential A Theory of Justice (Revised, 1999) the two concepts have been linked in popular discourse, according to the principle ‘justice as fairness’.
As fairness, justice is cited frequently, for example, in academic codes, as fairness to students, including especially refraining from exploiting free academic labour, and ensuring credit is given for any academic work they may have depended on (SFU, 1992) and viewing academics “as role models (who) must follow a professional code of ethics” to ensure “students receive a fair, honest and uncompromising education” from teachers who “demonstrate integrity, impartiality and ethical behavior” (Guyana, 2017).
Even viewed as ‘fairness’, however, ambiguities remain. As the Belmont Report notes. The idea of justice, “in the sense of ‘fairness in distribution’ or ‘what is deserved’” can be viewed from numerous perspectives, each of which needs to be considered, specifically, “(1) to each person an equal share, (2) to each person according to individual need, (3) to each person according to individual effort, (4) to each person according to societal contribution, and (5) to each person according to merit.” The authors also note that exposing a disadvantaged group to risk is an injustice (DHEW, 1978:6-7).
Fairness is also viewed as impartiality, an avoidance of bias or arbitrary ruling. In journalism, for example, “the primary value is to describe the news impartially – “without fear or favour”, as stated by New York Times “patriarch” Adolph Ochs (NYT, 2018). Similarly, the High-Level Expert Group on Artificial Intelligence (AI HLEG, 2019) endorses “diversity, non-discrimination and fairness – including the avoidance of unfair bias, accessibility and universal design, and stakeholder participation.” And the European University Institute opposesd acts that are arbitrary, biased or exploitative (EUI, 2019).
Justice, sometimes coined as ‘natural justice’ (CPA, 2017:11), can also be depicted in terms of rights (Stevens & Silbey, 2014; Asilomar, 2017; Access Now, 2018). That is how it appears in the Asilomar declaration. The principles themselves reflect a broadly progressive social agenda, “compatible with ideals of human dignity, rights, freedoms, and cultural diversity,” recognizing the need for personal privacy, individual liberty, and also the idea that “AI technologies should benefit and empower as many people as possible” and “the economic prosperity created by AI should be shared broadly, to benefit all of humanity.”
This interpretation of justice is also expressed as an endorsement of diversity and prohibition of discrimination (Sullivan-Marx, 2020; Brandom, 2018; CPA, 2017:11; BACB, 2014; etc.) based on various social, economic, cultural and other factors (this list varies from code to code). The National Union of Journalists code, for example, states explicitly that journalists should produce “no material likely to lead to hatred or discrimination on the grounds of a person’s age, gender, race, colour, creed, legal status, disability, marital status, or sexual orientation” (NUJ, 2011).
Justice, viewed from either the perspective of fairness or rights, can be expanded to include redress for current or past wrongs, or to prevent future wrongs. As early as 1973, U.S. Department of Health, Education and Welfare, on observing abuses in data collection, proposed a ‘Code of Fair Information Practice’. The intent of the code was to redress this imbalance and provide some leverage for individuals about whom data is being collected. The Toronto Declaration similarly calls for “clear, independent, and visible processes for redress following adverse individual or societal effects” (Brandom, 2018).
Depending on one’s perspective, the principle of justice may be listed together with, or apart from, any number of other principles, including fairness, rights, non-discrimination, and redress. That we have listed them here in one section does not presuppose that we are describing a single coherent core value or principle; rather, what we have here is a family of related and sometimes inconsistent principles that are often listed in the popular discourse as a single word, such as ‘justice’, as though there is some shared understanding of this.
Accountability and Explicability
The principles of accountability and explicability arise differently in computing and AI codes than it does in other ethical codes. In the case of academic and medical research, accountability is typically delegated to a process undertaken by a research ethics board (REB). Similarly, the Information and Privacy Commissioner of Ontario asserts that compliance with privacy rules and restrictions should be subject to independent scrutiny and that “the state must remain transparent and accountable for its use of intrusive powers through subsequent, timely, and independent scrutiny of their use” (Cavoukian, 2013).
In other disciplines, a range of additional processes describe practices such as predictability, auditing and review (Raden, 2019: 9). As the U.S. Department of Health and Welfare argued, data should only be used for the purposes for which it was collected. And this information, however used, should be accurate; there needs to be a way for individuals to correct or amend a record of identifiable information about themselves, and organizations must assure the reliability of the data and prevent misuse of the data. These, write the authors, “define minimum standards of fair information practice” (Ware, et.al., 1973:xxi).
In digital technology, accountability also raises unique challenges. The AI4People code, for example, adds a fifth principle to the four described by Beauchamp & Childress (1992), “explicability, understood as incorporating both intelligibility and accountability” where we should be able to obtain “a factual, direct, and clear explanation of the decision-making process” (Floridi et al. 2018). As (Fjeld, 2020) summarizes, “mechanisms must be in place to ensure AI systems are accountable, and remedies must be in place to fix problems when they’re not.” Also, “AI systems should be designed and implemented to allow oversight.”
Finally, says Fjeld, “important decisions should remain under human review.” Or as Robbins (2019) says, ‘Meaningful human control’ is now being used to describe an ideal that all AI should achieve if it is going to operate in morally sensitive contexts.” As Robbins argues, “we must ensure that the decisions are not based on inappropriate considerations. If a predictive policing algorithm labels people as criminals and uses their skin color as an important consideration then we should not be using that algorithm.”
Many of the codes of ethics, especially those dedicated to research, express openness as a core value, though often with conditions attached. The Sorbonne Declaration, for example, states “research data should, as much as possible be shared openly and reused, without compromising national security, institutional autonomy, privacy, indigenous rights and the protection of intellectual property” (Sorbonne Declaration, 2020). Similarly, the Declaration of Helsinki states “researchers have a duty to make publicly available the results of their research on human subjects and are accountable for the completeness and accuracy of their reports” (WMA, 2013).
Another project, FAIRsFAIR, is based on the the FAIR Guiding Principles (GoFAIR, 2020) for scientific data management and stewardship (Wilkenson, et.al., 2016). The principles (and the acronym derived from them) are “Findability, Accessibility, Interoperability, and Reusability—that serve to guide data producers and publishers as they navigate around these obstacles, thereby helping to maximize the added-value gained by contemporary, formal scholarly digital publishing.”
In many cases, openness is described in terms of access serving the public good. The Asilomar Convention includes a principle of openness representing learning and scientific inquiry as “public goods essential for well-functioning democracies” (Stevens & Silbey, 2014). Citing The Research Data Alliance’s 2014 “The Data Harvest Report” the Concordat Working Group, (2016) authors write “the storing, sharing and re-use of scientific data on a massive scale will stimulate great new sources of wealth” (Genova, et.al., 2014).
Openness is also described in some principles as openness of access to services. The IFLA (2019), for example, expresses “support for the principles of open access, open source, and open licenses” and “provision of services free of cost to the user.” The Canadian Nurses association code includes “advocating for publicly administered health systems that ensure accessibility, universality, portability and comprehensiveness in necessary health-care services” (CAN, 2017).
Openness is also described in some principles as ‘transparency’ of methods and processes (IA, 2019; Raden, 2019: 9; Cavoukian, 2013; CSPL, 1995) in a way that often references accountability (as referenced above). The Accenture code, for example, urges professionals to foster transparency and accountability (Accenture, 2016:5). The High-Level Expert Group on Artificial Intelligence (AI HLEG) also advocates transparency, which includes traceability, explainability and communication.
Finally, openness can be thought of as the opposite of secrecy, as mentioned in the Department of Health, Education and Welfare report, stating that individuals should have a way to find out what information about them is in a record and how it is used (Ware, et.al., 1973). It is also the opposite of censorship (IFLA, 2019; ALA, 2008).
Common Cause / Solidarity
Many codes of ethics also explicitly endorse an advocacy role for professionals to promote the values stated in the code. The AMA Declaration of Professional Responsibility, for example, asserts a commitment to “advocate for social, economic, educational, and political changes that ameliorate suffering and contribute to human well-being” (Riddick, 2003).
The codes vary from advice to “teach what uplifts and unites people and refuse to be, in any way whatsoever, the propagandists of a partisan conception” (Soleil, 1923) to establishing a shared vision of teaching and to “to identify the values, knowledge and skills that are distinctive to the teaching profession” (OCT, 2016) to expressing solidarity with other members of the profession, for example, stating that criticism of other members will be conducted in private (BCTF, 2020).
Obligations and Duties
As Feffer (2017) observes, our duties often conflict. For example, we may read, “As a representative of the company, you have one set of responsibilities. As a concerned private citizen, you have other responsibilities. It’s nice when those converge, but that’s not always the case.”
We might think, for example, that a practitioner always has a primary duty to their client. Thus a doctor, lawyer, or other professional tends to the interests of the client first. A look at practice, however, makes it clear this is not the case. A doctor may (in some countries) refuse to perform a service if a patient cannot pay. An educator may be required to report on a student’s substance abuse problem or immigration status.
And often, the locus of duty is not clear. For example, if a company is skewing the data used in order to sway a model toward a particular set of outcomes, does an employee have a duty to disclose this fact to the media? There may be some cases where a company is legally liable for the quality of its analytics, while in other cases (such as marketing and promotion) the requirement is less clear.
If we widen our consideration beyond simple transactions, the scope of our duties widens as well. Our duty to travel to Africa to support a learning program may not conflict with a duty to preserve the environment for people who have not yet been born. (Saugstad, 1994; Wilkinson & Doolabh, 2017) Or our desire to eat meat may conflict with what activists like Peter Singer might consider a duty to animals (Singer, 1979).
This this section we look briefly at the different entities to which different code argue that we owe allegiance, loyalty, or some other sort of obligation or duty.
Most ethical codes abnegate serving or benefitting oneself, and where the self is concerned, it is typically in the service of the wider ethic, for example, our obligations as role models (Guyana, 2017). The Nolan principles, for example, make clear that the ethics of a member of the public service is selflessness (CSPL, 1995), though there is occasional acknowledgement of a duty to self (AMA, 2001).
And yet, many of the ethical principles described in the code could be construed as the cultivation of a better self, for example, one who is honest, trustworthy, integral, objective and open (this list varies from code to code) (IMIA, 2015; CSPL, 1995; CPA, 2017; IA, 2017; AITP, 2017; etc.) as well as “self-knowledge regarding how their own values, attitudes, experiences, and social contexts influence their actions, interpretations, choices, and recommendations” (IMIA, 2015).
And some principles might be thought of as promoting some desirable attributes of self, even if referring to these in others: autonomous self-realisation, human agency, and individual capabilities, for example (Floridi, et.al., 2018:7), or to “participate in programmes of professional growth like in-service education and training, seminars, symposia workshops, conferences, self study etc.” (Mizoram, 2020).
We included a place-holder for duties or obligations to the less fortunate because of an earlier reference to Peter Singer’s (2009) The Life You Can Save. Statements of any obligation toward the poor or less fortunate are impossible to find in any of the ethical codes, however, with the exception of references to specific clients of a profession, as discussed below).
That is not to say that the less fortunate are completely omitted from ethical codes. As far back as Hammurabi’s Code is the edict, “the strong may not oppress the weak” (Gilman, 2005:4n3). At the same time, the resistance to considering such matters is telling, as summarized here: “Advocates have urged that considerations for the poor, illegal immigrants, rain forests, tribal rights, circumcision of women, water quality, air quality and the right to sanitary facilities be put into codes for administrators. As important as these issues might be they distort the purpose of ethics codes to the point that they are confusing and put political leadership in the position of quietly undermining them” (Ibid:47).
Ethical codes for teachers or academics often specify obligations or duties to students, though in different ways. For example, Le code Soleil assigns a three-fold responsibility to teachers: to train the individual, the worker, and the citizen. Education, according to the code, “is the means to give all children, whatever their diversity, to reach their maximum potential” (Soleil, 1923). The National Education Association code urges teachers to “strive to help each student realize his or her potential as a worthy and effective member of society” (NEA, 1975). Further, the Open University code asserts that “students should be engaged as active agents in the implementation of learning analytics (e.g. informed consent, personalised learning paths, interventions” (OU, 2014:4.3.2).
Parent or Guardian, Children
Parents stand in two roles in codes of ethics. The first is to act as a proxy for children with respect to matters of consent (Kay et al. 2012). The second is as special interests that need to be protected; for example, an Indian code of ethics advises teachers to “refrain from doing any thing which may undermine students confidence in their parents or guardians” (NCERT, 20910; Mizoram, 2020) and with whom teachers need to maintain an open and trusting relationship (OCT, 2020).
Data collection began early in the field of digital media, with the FTC noting that “The practice is widespread and includes the collection of personal information from even very young children without any parental involvement or awareness” (Ibid:5) It is worth noting that the principles are designed specifically to protect consumers, and that they are addressed specifically toward industry. (Pitofsky, et.al., 1998:ii)
In the IEEE code there is a detailed section on ‘working with children’ that contains provisions on safety and security, confidentiality, and whistle-blowing, noting specifically that “Adults have a responsibility to ensure that this unequal balance of power is not used for their personal advantage” (IEEE, 2017). Finally, “the Information Technology Industry Council has joined the conversation around children’s rights with a focus on emerging technologies, publishing a list of principles to guide the ethical development of artificial intelligence (AI) systems” (UC Berkeley, 2019).
In many ethical codes the first and often only duty is to the client. This is especially the case for service professions such as finance and accounting, legal representation, where this is expressed as fiduciary duties, which are “special obligations between one party, often with power or the ability to exercise discretion that impacts on the other party, who may be vulnerable” (Wagner Sidlofsky, 2020).
In health care the needs of the client are often paramount. For example, the Declaration of Helsinki (WMA, 2013) states ‘The health of my patient will be my first consideration,’” and cites the International Code of Medical Ethics in saying, “A physician shall act in the patient’s best interest when providing medical care.” It is thus “the duty of the physician to promote and safeguard the health, well-being and rights of patients, including those who are involved in medical research” (Ibid). In cases where multiple duties are owed, the client may be assigned priority, as in the case of medical research codes. “When research and clinical needs conflict, prioritize the welfare of the client” (BACB, 2014).
There is ambiguity in the concept of client, particularly with respect to the idea that the duty is to the client because the client is the one paying the bills. When care is paid by insurance, or through government programs, or corporate employers, the service recipient and the payer may be two distinct. Similarly, in digital media, costs may be paid by advertisers or publishers, who may then assert moral priority. (Done, 2010). However, as Luban (2018:187) argues, “’who pays the whistler calls the tune’ is not a defensible moral principle.”
Research ethics codes commonly describe a duty of the researcher to the research subject, beginning with the Nuremberg Principles and established throughout the practice thereafter. The rresponsibilities to research participants include informed consent, transparency, right to withdraw, reasonableness of incentives, avoidance and mitigation of harm arising from participation in research, and privacy (BERA, 2018).
In the field of data research and analytics this principle is often retained. Accenture’s universal principles for data ethics, for example, state that the highest priority is “the person behind the data” (Accenture, 2016:5). Similarly, the Insights Association code (2019) states “respect the data subjects and their rights.” In journalism, asa well, “ethical journalism treats sources, subjects, colleagues and members of the public as human beings deserving of respect” (SPJ, 2014).
Employer or Funder
Public service employees are not surprisingly obligated to their employer. “Members of the public service… are tasked with “loyally carrying out the lawful decisions of their leaders and supporting ministers in their accountability to Parliament and Canadians” (TBS,2011:1.1-1.2)
The same sometimes holds true in the case of ethical codes for teachers. They may be required to “cooperate with the head of the institution and colleagues in and outside the institution in both curricular and co-curricular activities” and that a teacher should “recognize the management as the prime source of his sustainable development” (Mizoram, 2020) or to “abide by the rules and regulations established for the orderly conduct of the affairs of the University” (SFU, 1992).
The same may apply for employees in the private sector. Information technology professionals, for example, may be asked “to guard my employer’s interests, and to advise him or her wisely and honestly” (AITP, 2017). Journalists, as well, are subject to obligations to the newspaper (NUJ., 2936). Even funders may make a claim on the duties of the researcher (Dingwell, et.al., 2017).
Colleagues, Union or Profession
Professional associations and unions frequently include loyalty to the professional association or union as a part of the code of ethics, either explicitly, or expressed as an obligation owed to colleagues (NUJ, 1936; AITP, 2017; SFL, 1992; NEA, 1975; etc.). This is related to the idea that members are forming a voluntary association. “If a member freely declares (or professes) herself to be part of a profession, she is voluntarily implying that she will follow these special moral codes. If the majority of members of a profession follow the standards, the profession will have a good reputation and members will generally benefit” (Weil, 2008).
The term ‘stakeholders’ is sometimes used without elaboration to indicate the presence of a general duty or obligation
(BERA, 2018). Fjeld (2020) asserts for example that “developers of AI systems should make sure to consult all stakeholders in the system and plan for long-term effects.” The Open University policy is based on “significant consultation with key stakeholders and review of existing practice in other higher education institutions and detailed in the literature” (OU, 2014:1.2.6). Similarly, one of the DELICATE principles (Drachsler & Greller, 2016) requires researchers “talk to stakeholders and give assurances about the data distribution and use.”
What is a stakeholder? It expands on the concept of ‘stockholder’ and is intended to represent a wider body of interests to which a company’s management ought to be obligated (SRI, 1963). Freeman (1984:25) defines it as “any group or individual who can affect, or is affected by, the achievement of a corporation’s… or organization’s purpose… or performance”. He bases it on “the interconnected relationships between a business and its customers, suppliers, employees, investors, communities and others who have a stake in the organization” (Ledecky, 2020). There are many definitions of ‘stakeholder’ (Miles,2017:29) and no principled way to choose between them.
Publishers and Content Producers
Librarians are subject to special obligations to publishers, according to some codes. For example, “Librarians and other information workers’ interest is to provide the best possible access for library users to information and ideas in any media or format, whilst recognising that they are partners of authors, publishers and other creators of copyright protected works” (IFLA, 2012).
This responsibility is extended in other fields as a prohibition against plagiarism (EUI, 2019; BACB, 2014; SPJ, 2014; NUJ, 2011; NYT, 2017; etc.) and taking credit for the work of others (AITP, 2017; IEEE, 2020; BACB, 2014; etc.).
References to a responsibility to society are scarce, but they do exist. BERA (2018) argues for a responsibility to serve the public interest, and in particular, responsibilities for publication and dissemination. The ‘Nolan principles’, (CSPL, 1995) state “Holders of public office are accountable to the public for their decisions and actions and must submit themselves to the scrutiny necessary to ensure this.”
In the field of data analytics, the last two of the Computer Ethics Institute’ Ten Commandments’ recommend computer professionals “think about the social consequences” and to “ensure consideration and respect for other humans” (CEI, 1992). Though as Metcalf (2014) notes, “it appears to be the only computing ethics code that requires members to proactively consider the broad societal consequences of their programming activities” (my italics). Subsequently, the Royal Society (Drew, 2016) recommended data scientists “be alert to public perceptions.”
Law and Country
Although it has been established that there is not an ethical duty to obey an unethical law, a number of ethical codes nonetheless include respect for the law in one way or another, for example, in reporting child protection issues (BCTF, 2020), compliance with law as an ‘overarching principle’ (IA, 2019), or “operate within the legal frameworks (and) refer to the essential legislation (Drachsler & Greller, 2016).
Meanwhile, the Association of Information Technology Professionals Code of Ethics asserts “I shall uphold my nation and shall honor the chosen way of life of my fellow citizens,” though it is no longer extant and as Metcalf (2016) comments, “it is decades old and has some anachronisms that clash with globalized ethos of computing today.” Despite this, it was cited (in EDUCAUSE Review) as recently as 2017 (Woo, 2017).
The environment is rarely mentioned in ethical codes, though it appears in a statement of obligations to “society, its members, and the environment surrounding them” (ACM, 2018) and as “societal and environmental wellbeing – including sustainability and environmental friendliness, social impact, society and democracy” (AI HLEG, 2019).
Bases for Values and Principles
What grounds these codes of ethics? On what basis do their authors assert that this code of ethics, as opposed to some hypothetical alternative, is the code of ethics to follow? A typical explanation might be that “An individual’s professional obligations are derived from the profession and its code, tradition, society’s expectations, contracts, laws, and rules of ordinary morality” (Weil, 2008), but a closer examination raises as many questions as it answers.
Many codes simply assert that the principles embodied in the code are universal principles. Universality may be seen as a justification for moral and ethical principles; if the principle is believed by everyone, then arguably it should be believed here.
For example, the Universal Declaration of Ethical Principles for Psychologists asserts, “The Universal Declaration describes those ethical principles that are based on shared human values” (IUPSYS, 2008). It later asserts “Respect for the dignity of persons is the most fundamental and universally found ethical principle across geographical and cultural boundaries, and across professional disciplines” (Ibid). So we see here universality being asserted as a foundation underlying a set of ethical principles. Similarly, the Asolomar Convention states that “Virtually all modern societies have strong traditions for protecting individuals in their interactions with large organizations… Norms of individual consent, privacy, and autonomy, for example, must be more vigilantly protected as the environments in which their holders reside are transformed by technology” (Stevens & Silbey, 2014).
Additional studies, such as Fjeld, et.al. (2020) that suggest that we have reached a consensus on ethics and analytics. We argue that this is far from the case. The appearance of ‘consensus’ is misleading. For example, in the Fjeld, et.al., survey, though 97% of the studies cite ‘privacy’ as a principle, consensus is much smaller if we look at it in detail (Ibid:21), and the same if we look at the others, eg. Accountability (Ibid:28). Aassertions of universality made elsewhere (for example: Pitofsky,1998:7; Singer & Vinson, 2002; CPA, 2017; Raden, 2019: 11) can be subject to similar criticisms.
In their examination of teacher codes of ethics, Maxwell and Schwimmer (2016) found “analysis did not reveal an overlapping consensus on teachers’ ethical obligations.” Nor are they alone in their findings; citing Campbell (2008:358) they observe that “despite extensive research on the ethical dimensions of teaching, scholars in the field do not appear to be any closer to agreement on ‘the moral essence of teacher professionalism’.” Similarly, Wilkinson (2007:382) “argues that the teaching profession has failed ‘to unite around any agreed set of transcendental values which it might serve’.” And van Nuland & Khandelwal (2006:18) report “The model used for the codes varies greatly from country to country.” The selection below is a sample; many more codes may be viewed in the EITCO website (IIEP, 2020).
The High-Level Expert Group on Artificial Intelligence cites four ethical principles, “rooted in fundamental rights, which must be respected in order to ensure that AI systems are developed, deployed and used in a trustworthy manner” (AI HLEG, 2019).
As noted above, the Access Now report specifically adopts a human rights framework “The use of international human rights law and its well-developed standards and institutions to examine artificial intelligence systems can contribute to the conversations already happening, and provide a universal vocabulary and forums established to address power differentials” (Access Now, 2018:6).
The Toronto Declaration “focuses on the obligation to prevent machine learning systems from discriminating, and in some cases violating, existing human rights law. The declaration was announced as part of the RightsCon conference, an annual gathering of digital and human rights groups” (Brandom, 2018).
Nonetheless, it is not clear what these fundamental rights are. Their statement in documents such as the U.S. Bill of Rights, the Canadian Charter of Rights and Freedoms, or the Universal Declaration of Human Rights, is very different. Is the right to bear arms a fundamental right? Is the right to an education a fundamental right?
Arguments drawing from statements of fact about the world are sometimes used to support ethical principles. For example, the Universal Declaration of Ethical Principles for Psychologists asserts, “All human beings, as well as being individuals, are interdependent social beings that are born into, live in, and are a part of the history and ongoing evolution of their peoples… as such, respect for the dignity of persons includes moral consideration of and respect for the dignity of peoples” (IUPSYS, 2008).
Against such assertions of fact the “is-ought” problem may be raised. As David Hume (1739) argued, moral arguments frequently infer from what ‘is’ the case to what ‘ought’ to be the case, but “as this ought, or ought not, expresses some new relation or affirmation, ’tis necessary that it should be observed and explained; and at the same time that a reason should be given” (Hume, 1888:469). Such ‘oughts’ may be supported with reference to goals or requirements (see below), or with reference to institutional facts, such as laws (Searle, 1964).
Balancing Risks and Benefits
The AI4People declaration states “An ethical framework for AI must be designed to maximise these opportunities and minimise the related risks” (Floriodi, et.al., 2018:7). Similarly the Concordat Working Group (2016) document is of open data with the need to manage access “in order to maintain confidentiality, protect individuals’ privacy, respect consent terms, as well as managing security or other risks.” And the AI4People starts from the premise that “an ethical framework for AI must be designed to maximise these opportunities and minimise the related risks” (Floridi, et.al., 2018:7).
The balancing of risks and benefits is a broadly consequentialist approach to ethics and therefore results in a different calculation in each application. For example, the balancing of risk and benefit found in the Common Rule is focused more specifically on biomedical research, and it has to be asked, is biomedicine the ethical baseline? “Not all research has the same risks and norms as biomedicine… there has remained a low-simmering conflict between social scientists and IRBs. This sets the stage for debates over regulating research involving big data.” (Metcalfe, 2016)
It also requires an understanding of what the consequences actually are. Four of the five principles recommended by the House of Lords Select Committee on AI represent a consequentialist approach (Clement-Jones, et.al, 2018: para 417). But what are those consequences? The Committee quotes the Information Commissioner’s Office (ICO) as stating that there was a “need to be realistic about the public’s ability to understand in detail how the technology works”, and it would be better to focus on “the consequences of AI, rather than on the way it works”, in a way that empowers individuals to exercise their rights (Ibid: para 51), but this may be unrealistic.
And perhaps ethics isn’t really a case of balancing competing interests. The Information and Privacy Commissioner in Ontario (Cavoukian, 2013) asserts that “a positive-sum approach to designing a regulatory framework governing state surveillance can avoid false dichotomies and unnecessary trade-offs, demonstrating that it is indeed possible to have both public safety and personal privacy. We can and must have both effective law enforcement and rigorous privacy protections.”
Requirements of the Profession
A requirement is a statement about what a person must believe, be or do in order to accomplish a certain objective or goal. For example, the Universal Declaration of Ethical Principles for Psychologists asserts, “competent caring for the well-being of persons and peoples involves working for their benefit and, above all, doing no harm… (it) requires the application of knowledge and skills that are appropriate for the nature of a situation as well as the social and cultural context” (IUPSYS, 2008). Similarly, the American Library Association sees its role as requiring “a special obligation to ensure the free flow of information and ideas to present and future generations” (ALA, 2008). The IFLA similarly argues that “librarianship is, in its very essence, an ethical activity embodying a value-rich approach to professional work with information” (IFLA, 2012).
The same document also later asserts that “Integrity is vital to the advancement of scientific knowledge and to the maintenance of public confidence in the discipline of psychology,” which is the same type of argument, however, the objectives are much less clearly moral principles: the “advancement of scientific knowledge” and “the maintenance of public confidence.” Such arguments often proceed through a chain of requirements; IUPSYS (2008) continues, for example, to argue that “Integrity is based on honesty, and on truthful, open and accurate communications.”
Such principles may be expressed in two ways: either derived, or conditional. The principle is derived if the antecedent is already an ethical principle. In the first IUPSYS example above, for example, “competent caring for the well-being of persons and peoples” may have been previously established as an ethical principle, from which the derived principle ‘working for their benefit’ is also established. The principle may be expressed as a conditional that describes what is entailed on (say) joining a profession: if one is engaged in competent caring for the well-being of persons and peoples then this requires working for their benefit.
Against such assertions of requirements, several objections may be brought forward. The first method is to argue that the requirement does not actually follow from the antecedent; one might argue, for example that competent caring does not entail working for the person’s benefit; it may only involve following proper procedures without regard to the person’s benefit. Additionally, one might argue that the antecedent has not in fact been established; for example, one might argue that being a psychologist doesn’t involve caring at all, and might only involve addressing certain disruptions in human behaviour. A criminal psychologist might take this stance, for example.
Social Good or Social Order
Social good, however defined, may be the basis of some ethical principles. The preamble to the Society for Professional Journalists (SPJ) code of ethics states that the primary function of journalism, according to the statements, is to inform the public and to serve the truth, because “public enlightenment is the forerunner of justice and the foundation of democracy” (SPJ, 2014).
A basis in social order, however, invites relativism. People’s ethical judgements are relative (Drew, 2016). “People’s support is highly context driven. People consider acceptability on a case-by-case basis, ﬁrst thinking about the overall policy goals and likely intended outcome, and then weighing up privacy and unintended consequences” (Ibid). This relativism is clear in a statement from a participant: “Better that a few innocent people are a bit cross at being stopped, than a terrorist incident – because lives are at risk.” And this relativism often reflects their own interests: “a direct personal beneﬁt (e.g. giving personalized employment advice), beneﬁt to a local community, or public protection” (Ibid).
‘Social order’ can be construed to mean national interest. We see this in ethics statements guiding public service agencies and professionals. For example, Russell T. Vought, issued a memo asserting that “Office of Management and Budget (OMB) guidance on these matters seeks to support the U.S. approach to free markets, federalism, and good regulatory practices (GRPs), which has led to a robust innovation ecosystem” (Vought, 2020). The resulting ‘Principles for the Stewardship of AI Applications’ included such things as public participation, public trust, and scientific integrity, but also included risk assessment and management along with benefits and costs. The document also urged a non-regulatory approach to ethics in AI. A different society might describe ethics in government very differently.
A principle of ‘fairness’ is frequently cited with no additional support or justification.
Often, fairness is defined as essential to the ethics of the profession. The New York Times, for example, “treats its readers as fairly and openly as possible” and also “treats news sources just as fairly and openly as it treats readers” (NYT, 2018).
Fairness may be equated with objectivity. For example, a journalist may say, “it is essential that we preserve a professional detachment, free of any whiff of bias” (NYT, 2018).
While acknowledging that “there is nothing inherently unfair in trading some measure of privacy for a benefit,” the authors of a 1973 report for the U.S. Department of Health, Education and Welfare addressing the then nascent practice of electronic data management noted that “under current law, a person’s privacy is poorly protected against arbitrary or abusive record-keeping practices” (Ware, et.al., 1973). Hence they proposed what they called a ‘Code of Fair Information Practice’.
The advancement of knowledge and learning is often considered to be in and of itself a moral good. For example, it is used in the Universal Declaration of Ethical Principles for Psychologists to justify the principle of integrity: “Integrity is vital to the advancement of scientific knowledge and to the maintenance of public confidence in the discipline of psychology” (IUPSYS, 2008). Epistemological justification is also found in journalistic ethics: “relationships with sources require the utmost in sound judgment and self discipline to prevent the fact or appearance of partiality” (NYT, 2018). And in the case of AI ethics, it may be simply pragmatic: “our ‘decision about who should decide’ must be informed by knowledge of how AI would act instead of us” (Floridi, et.al., 2028:21).
Against this argument, one may simply deny that knowledge and learning are moral goods, and are simply things that people do, and can often be harmful (as in “curiosity killed the cat”). More often, we see such responses couched in specific terms, asserting that seeking some particular knowledge is not inherently good, for example, knowledge related to advanced weapons research, violations of personal confidentiality, and a host of other real or imagined harms. Seneca, for example, argued “This desire to know more than is sufficient is a sort of intemperance” (Letter 88:36).
In order to do any number of things, you need trust, or some of the components of trust. As a result, the elements of trust in themselves can be cited as justification for moral principles. For example, the Universal Declaration of Ethical Principles for Psychologists writes “Integrity is vital… to the maintenance of public confidence in the discipline of psychology” (IUPSYS, 2008). Chartered Financial Analysts seek to “promote the integrity and viability of the global capital markets for the ultimate benefit of society” (CFA, 2019).
Similar principles underlie ethics in journalism; “integrity is the cornerstone of a journalist’s credibility” (SPJ, 1996). Similarly, the New York Times asserts, “The reputation of The Times rests upon such perceptions, and so do the professional reputations of its staff members.” If we here interpret ‘public confidence’ as an aspect of trust, we see how the authors are appealing to the principle of trust to support the assertion that integrity is a moral principle.
Against this it may be argued that trust is neither good not bad in and of itself, and indeed, that trust may be abused in certain cases, which could make measures that promote trust also bad. Moreover, it could be argued that trust is too fragile a foundation for moral; principles, as it may be broken even without ill attempts. Further, it may be argued that trustless systems are in fact morally superior, because they do not create the possibility that trust may be breached, thus preserving the integrity of whatever it was that trust was intended to support.
Another way to define an ethical principle’ is to say that it is descriptive of ‘conduct that you (or your organization) would be willing to defend’. For example, the National Union of Journalist code of conduct (NUJ, 2011) offers “guidance and financial support of members who may suffer loss of work for conforming to union principles.”
“Through years of courageous struggle for better wages and working conditions its pioneers and their successors have kept these aims in mind, and have made provision in union rules not only for penalties on offenders, but for the guidance and financial support of members who may suffer loss of work for conforming to union principles” (NUJ, 1936).
Includes burden or onus – responding to U.S. Whitehouse – Guidance for Regulation of Artificial Intelligence Applications – Responding to these guidelines, the American Academy of Nursing argued for a less business-focused assessment of the risks and benefits of AI, saying “federal agencies should broaden the concept around use of AI related social goals when considering fairness and non-discrimination in healthcare.” They also urged that “federal agencies consider patient, provider, and system burden in the evaluation of AI benefits and costs” and “include data accuracy, validity, and reliability” in this assessment (Sullivan-Marx, 2020)
Results of the Study
After having studied a certain number of codes of ethics, in the light of the applications of analytics and arising ethical issues considered above, the following statements can be asserted.
- None of the statements address all of the issues in learning analytics extant in the literature, and arguably, all of these statements, taken together, still fail to address all these issues.
- Those issues that they address, they often fail to actually resolve. Often the principles state what should be considered, but leave open what should be the resolution of that consideration.
- There are legal aspects to analytics, and there are ethical aspects, and there is a distinction between the two, though this distinction is not always clear.
- Although there is convergence around some topics of interest, there is no consensus with respect to the ethics involved.
- In fact, there are conflicts, both between the different statements of principles, and often, between the principles themselves (often described as a need to ‘balance’ competing principles).
- Even were there consensus, it is clear that this would be a minimal consensus, and that important areas of concern addressed in one domain might be entirely overlooked in another domain.
- Ethical principles and their application vary from discipline to discipline, and from culture to culture.
- There is no common shared foundation for the ethical principles described. As we will see below, these statements of principles select on an ad hoc basis from different ethical ideas and traditions.
- Often these principles include elements of monitoring and enforcement, thus begging the question of why or for what reason an individual would adhere to the ethical principle stated.
It is premature (if it is possible at all) to talk about “the ethics of such and such” as though we have solved ethics. There are multiple perspectives on ethics, and these are represented in the very different ethical codes from various disciplines. Approaches based in simple principles, such as an appeal to consequences, or such as in terms of rights and duties, and as such, as statements of rules or principles, fail to address the complexity of ethics especially as regards learning and analytics. The assertion of a universal nature of ethics doesn’t take into account context and particular situations, and it doesn’t take into account larger interconnected environment in which all this takes place.
Additionally, based in simple principles don’t take into account how analytics themselves work. Analytics systems are not based on rules or principles, they are statistical, using techniques such as clustering and regression. As such, their input is going to be complex, and they will produce unexpected consequences in a way that reflects the complexity of humans and human society.
There is an argument, with which we are sympathetic, that when we ask ethical questions, such as “what makes so-and-so think it would be appropriate to post such-and-such?” we are not looking for a single answer, but a complex of factors based on individual identity, society, circumstances and perspective. This suggests an ethics based on different objectives – not ‘rights’ or ‘fairness’ but rather things like a sense of compassion or on a philosophical perspective that uses a relational and context-bound approach toward morality and decision making, for example, as found in work based in conviviality or the ethics of care.